Critical analysis ofCMIPspast climate model projections in a regional context: The Iberian climate

Abstract

AbstractThe Iberian Peninsula is a known climate change hotspot. In the last decades, the Coupled Model Intercomparison Project (CMIP) has allowed for thousands of Global Climate Model (GCM) simulations to be conducted, an important tool to assess and understand future changes in Earth's climate. The comparison of past future projections with observations provides a robust evaluation of the models’ skill according to different emission scenarios. Here, a comprehensive performance assessment of GCM‐simulated mean, maximum, and minimum temperatures, and accumulated precipitation is conducted for Iberia, by retrospectively comparing historical simulations and past future projections from CMIPs 1 to 6 with reference datasets. From means to extremes, and multi‐year intra‐annual cycles to inter‐annual trends, GCM simulations are compared with Iberia0.1 and E‐OBS observational gridded datasets, and ERA5 reanalysis. The matching between the variables’ distributions is assessed through the distribution added values (DAVs), a measure of gain or loss in performance between CMIPs. Results show relevant improvements in the description of the Iberian climate throughout the CMIP effort, for historical and past future periods. While the representation of intra‐annual cycles (inter‐annual trends) is enhanced after CMIP3 (CMIP2), GCMs from all CMIPs are generally able to depict the observed warming trend. Nevertheless, until 2021, a slight detrimental effect in the performance of CMIP6 models is found, in comparison with CMIP5 ones, with positive DAVs obtained only for past future temperature projections (less than 2%). A continuous monitorization of modelling accuracy for Iberia is needed, considering the increasing relevance of climate change information for adaptation strategies.